Deciphering signatures of mutational processes operative in human cancer.
Alexandrov LB, Nik-Zainal S, Wedge DC, Campbell PJ, Stratton MR. · 2013
Read Deciphering signatures of mutational processes operative in human cancer. by Alexandrov LB, Nik-Zainal S, Wedge DC, Campbell PJ, Stratton MR. (2013) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc3588146.txt.
Introduction
Ludmil Alexandrov and colleagues set out a method for reading the fingerprints of cancer-causing processes in Cell Reports in 2013. Every cancer genome carries thousands of mutations gathered over a lifetime, left by DNA damage and faulty repair. A small share drive the cancer; the rest are passengers. Global projects were cataloguing these mutations by the thousand, yet no framework existed for sorting the catalogue back into the separate processes that wrote it.
Treating the task as blind source separation, borrowed from signal processing, their framework teases apart the mixed mutation patterns into distinct signatures, one per underlying process. Each signature reflects the kinds of DNA change a process favors and the sequence letters around them. With the maths in place, catalogues from many cancers can be mined the same way. The authors present the approach as the doorway to learning how each cancer was made, process by process.
Catalog record
| CREATOR | Alexandrov LB, Nik-Zainal S, Wedge DC, Campbell PJ, Stratton MR. |
|---|---|
| YEAR | 2013 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Deciphering Signatures of Mutational Processes Operative in Human Cancer By Ludmil B Alexandrov, Serena Nik-Zainal, David C Wedge, Peter J Campbell, Michael R Stratton Published in: Cell Reports DOI: 10.1016/j.celrep.2012.12.008 ABSTRACT Summary The genome of a cancer cell carries somatic mutations that are the cumulative consequences of the DNA damage and repair processes operative during the cellular lineage between the fertilized egg and the cancer cell. Remarkably, these mutational processes are poorly characterized. Global sequencing initiatives are yielding catalogs of somatic mutations from thousands of cancers, thus providing the unique opportunity to decipher the signatures of mutational processes operative in human cancer. However, until now there have been no theoretical models describing the signatures of mutational processes operative in cancer genomes and no systematic computational approaches are available to decipher these mutational signatures. Here, by modeling mutational processes as a blind source separation problem, we introduce a computational framework that effectively addresses these questions. Our approach provides a basis for characterizing mutational signatures from cancer-derived somatic mutational catalogs, paving the way to insights into the pathogenetic mechanism underlying all cancers. INTRODUCTION All cancer genomes carry somatic mutations. A small minority are “drivers” of oncogenesis that confer selective clonal growth advantage. The remainder are “passengers” that have not been positively selected during the evolution of the neoplasm (Stratton et al., 2009). Global sequencing projects are generating catalogs of somatic mutations from tens of thousands of cancers (Hudson et al., 2010). The mutations within these catalogs are the cumulative result of all the somatic mutational mechanisms, including DNA damage and repair processes, that have been operative during the cellular lineage starting from the fertilized egg from which the cancer patient developed to the cancer cell (Stratton, 2011). Because the large majority of mutations in cancer genomes are believed to be passengers, their patterns are largely unmodified by selection (Rubin and Green, 2009). Thus, the mutational catalog from a cancer cell may be treated as a representative archaeological record bearing the combined imprints or signatures of the mutational processes that have been operative. Several mutational processes are already known to leave characteristic mutational signatures in the mutational catalogs of cancer cells. For example, analyses of mutated cancer genes in tumors of the lung and skin have shown that the classes of mutations found match those induced experimentally by tobacco carcinogens and ultraviolet light respectively, the major known exogenous carcinogenic influences in these cancer types (Hainaut and Pfeifer, 2001; Pfeifer et al., 2002, 2005). Notably, C:G > A:T transversions predominate in smoking-associated lung cancer, whereas C:G > T:A transitions occurring mainly at dipyrimidines and CC:GG > TT:AA double nucleotide substitutions are common in UV light-associated skin cancers. Thus, strong exposures to exogenous mutagens are known to leave their imprints as mutational patterns in cancer genomes. In principle, other biological processes may influence the patterns of somatic mutations found in human cancers. There may be additional exogenous mutagenic exposures. For example, many widely used chemotherapeutic cancer treatments are mutagens and some have already been shown to leave a distinctive mutational signature in the genomes of cancers recurring after therapy (Hunter et al., 2006). Moreover, there may be exogenous mutagenic exposures instrumental in the primary etiology of cancer that we are currently unaware of. Endogenous sources of mutagens may also contribute to mutations in cancer. For example, intrinsic cellular processes such as energy metabolism and lipid peroxidation are sources for reactive chemicals (e.g., reactive oxygen species) that cause DNA damage (Pluskota-Karwatka, 2008). These endogenous mutagens are known to generate certain subclasses of mutation and, thus, also might shape mutational catalogs within cancer genomes (Ames and Gold, 1991). The cell employs repair mechanisms that protect the integrity of the genetic code by alleviating and correcting the effects of exogenous and endogenous mutagens (Berwick and Vineis, 2000). For example, the base excision and nucleotide excision repair pathways act on DNA damage respectively caused by cellular metabolites and a wide variety of helix-distorting DNA lesions (Fuss and Cooper, 2006). These repair processes, in turn, influence the mutational signatures left by DNA damaging agents in the final catalog of mutations. This pertains when the repair processes are fully operative, for example the transcriptional strand bias observed in some mutational catalogs that is conferred by the transcription-coupled component of nucleotide excision repair (van Zeeland et al., 2005). It may also apply when they are malfunctioning, for example in the mutational signatures left by defective DNA mismatch repair (Greenman et al., 2007). In most human cancer types, the mutational signatures imprinted by DNA damage and repair processes have been subject to very limited characterization. In consequence, our understanding of the underlying mutational processes is poor. Previously, we reported an initial outline of a way to extract mutational signatures from the substitutions found in 21 breast cancer genomes (Nik-Zainal et al., 2012a). In this article, we provide a detailed description of our theoretical model that bridges the gap between mutational catalogs derived from cancer genomes and the mutational signatures contained in these catalogs. Further, we provide a systematic computational framework that can be used for accurately deciphering signatures of mutational processes from mutational catalogs of cancer genomes. We extensively evaluate our framework with simulated and real data, demonstrating that it allows incorporation of a wide variety of different mutation types (e.g., substitutions, indels, strand bias, kataegis, etc.). Our framework is freely available (see Experimental Procedures for details) and robust to a large range of different parameters that define its domain of applicability. Importantly, we demonstrate the applicability of the approach to genome and exome sequences and its potential to identify surprising biological insights. RESULTS Modeling Mutational Processes Operative in Cancer Genomes Somatic mutations are conventionally grouped into four classes; base substitutions, small indels, rearrangements, and copy number changes. These can be further subclassified into biologically meaningful subgroups. For example, base substitutions are often classified into six subtypes; C:G > A:T, C:G > G:C, C:G > T:A, T:A > A:T, T:A > C:G, and T:A > G:C. Classification of substitutions may be further refined by including the sequence context of each mutated base, either 5′ or 3′ or both. For example, a C:G > T:A mutation can be characterized as TpCpG > TpTpG (mutated base underlined and presented as the pyrimidine partner of the mutated base pair) generating 96 possible mutation types (6 types of substitution ∗ 4 types of 5′ base ∗ 4 types of 3′ base). This can be further elaborated by considering the transcriptional strand on which a substitution resides. In principle, similar approaches could be taken for the other major classes of mutation (i.e., indels, rearrangements, and copy number changes) and all classes and subclasses of mutation could be incorporated into one analysis. For the purpose of mathematical modeling, a limited number of features of a mutational signature need to be selected. The choice of features may be influenced by prior biological knowledge and is constrained by statistical considerations and the available data. In this study, a signature of a mutational process is represented as a discrete probability density function with a domain of preselected mutation features. Mathematically, mutation features can be expressed as a finite alphabet Ξ with K letters (each letter corresponds to a mutation feature) and, by